The key motivation of this study is to examine the application of autoregressive model for forecasting and trading the BDT/USD exchange rates from July 03, 2006 to April 30, 2010 as in-sample and May 01, 2010 to July 04, 2011 as out of sample data set. AR and ARMA models are benchmarked with a naïve strategy model. The major findings of this study is that in case of in-sample data set, the ARMA model, whereas in case of out-of-sample data set, both the ARMA and AR models jointly utperform other models for forecasting the BDT/USD exchange rate respectively in the context of statistical performance measures. As per trading performance, both the ARMA and naive strategy models outperform all other models in case of in-sample data set. On the other hand, both the AR and naive strategy models do better than all other models in case of out-of-sample data sets as per trading rformance.